By Jeremy Nighohossian
Wednesday, November 5, 2025
One of Robert F. Kennedy Jr.’s promises as secretary
of health and human services was to bring gold-standard science to our
public health regulators. He mentions it over and over again. But in his first six months, the quality of the
science he’s offering is closer to fool’s gold. And the problem is that what he
is doing could jeopardize, not help, public health.
When scientists refer to the gold standard of studies,
they mean a research approach that can isolate causal connections from the
other factors. The goal is to find out what causes a gain or loss for public
health and disregard other factors that may seem relevant but aren’t.
For example, imagine a study that found that people who
eat more fruits and vegetables live three years longer than those who don’t.
Readers might jump to the conclusion that these foods literally add years to
lives. And while such a study is correct — eating more fruits and vegetables is
good for you — it should not be taken in isolation. That would mean neglecting
all the other behaviors that people who eat more fruits and vegetables tend to
exhibit. They are more likely to exercise more and eat less food that’s
unhealthy, such as sugary treats. They’ve also probably eaten healthier for
many years. On the other hand, an 80-year-old who watches TV 16 hours a day and
eats three cheeseburgers for lunch and a pizza for dinner is not going to
increase his life expectancy by three years by adding an apple a day to his
diet.
The way scientists sort through data to determine how the
world works is to use randomization in studies. If researchers want to
understand the effect of fruits and vegetables, the ideal approach would be to
select a group of people at random to try to get a cross section of America.
Then, from this group, again randomly select a “control” group and an
“experimental” group. The researcher gives the experimental group the fruits
and vegetables and leaves the control group alone. This allows the researcher
to better isolate the change that was made and determine its effect.
That brings us to Secretary Kennedy. Instead of relying
on “gold standard” science, he often talks up “associative” studies. This
evokes the same kinds of studies common in the 1990s and 2000s when every other
week it seemed like coffee, eggs, and meat would flip back and forth between
being good for you and bad for you. (Here’s a primer on deciphering how these studies are reported by
the press.)
The 2025 Make America Healthy Again report put out by the Department
of Health and Human Services included over 500 citations, but only a few of
them seem to have been based on randomized controlled trials. The word
“randomized” doesn’t appear until page 23 in citation 103. The “What is Driving
the Increase in Childhood Chronic Disease?” section doesn’t include any direct
mention of randomized controlled trials. While such studies show a relationship
between one factor and an outcome, because many of the studies were not
randomized, they cannot definitely prove that the factor caused the
outcome.
Additionally, a major weakness of such associative
studies is that they often provide no mechanism to explain the relationship
— even an ex-post rationale. That is, there often is no explanation of how A
would lead to B — what processes take place inside the human body that would
explain how the factor in question caused the outcome of concern.
Of course, researchers can massage data in ways to
produce significant results even when none exist. In the mid-2010s, researchers were finding that many of the significant studies
of the past could not be replicated with the same approach but different
data, calling the studies’ results into question. The inability to reproduce
findings caused many to wonder how much of scientific research was being
distorted by a desire to get published. As a consequence of this “replication
crisis,” good researchers began what is called preregistering their approach.
By specifying exactly the analyses they intended to do before they had access
to data, they reduced the opportunity for manipulation.
Similarly, in the associative studies that Secretary
Kennedy and his team rely on, specifying a biological mechanism beforehand
would serve to reduce specious relationships. For example, if you looked for a
relationship between 100 random things and autism, chances are, you’d get
several false positives. Secretary Kennedy and his team are looking at these
random things and proclaiming a problem, when the reality is probably the
opposite. If you’re going to change national health policy after purporting to
find a relationship, a minimum requirement should be to explain how X leads to
Y.
In a recent announcement regarding Tylenol and autism, President Trump and RFK Jr. implored
pregnant women not to use acetaminophen except “in cases of extremely high
fever, that you feel you can’t tough it out” because a handful of studies
showed a connection. Meanwhile, they promoted a generic drug prescribed to
treat anemia and to counteract some of the extreme side effects of cancer
treatments as a treatment for autism, despite its not having gone through the standard set of clinical trials.
RFK Jr. promised that his Health and Human Services tenure would be built on gold-standard science, but the reality is more akin to cherry-picked science.
No comments:
Post a Comment